Enabling Concept-Based Relevance Feedback for Information Retrieval on the WWW
IEEE Transactions on Knowledge and Data Engineering
| Hi-index | 0.00 |
The huge information resources on the Internet have brought about many popular Internet knowledge discovery tools. A high level information gathering tool is necessary to cope with the diversity of information systems. Such a tool can release users from selecting the most appropriate tools among the various services. Also, since the database underlying any of the existing Internet services is only an incomplete snapshot of the whole Internet resources, the federation of several low level services can make maximum use of the resources. We present a system, ECHO, that gathers information from a particular kind of Internet knowledge discovery tools -- web search engines. It utilizes the unguided machine learning technique, data clustering, to classify the search results into semantic groups. Each of these cluster is a set of search results which are semantically related to each other. This technique is extremely effective when (1) user''s query is not specific or vague. (2) the number of search results is too large to sift through. However, the benefits gained from this kind of organization go beyond this effectiveness. As data clustering often leads to discovery, the system provides a consistent, integrated view of the gathered information. It can help users focus on the group of results of interest and find their real interests.